Administrative work rarely appears on a productivity dashboard, yet the time employees spend navigating routine processes can carry a significant organisational cost.
From chasing approvals and correcting expense reports to entering the same information across multiple systems, small administrative tasks can accumulate into hours of lost productive time. Brett Wheeldon, VP, Solution Advisory, APAC, SAP Concur, believes organisations need to look beyond the visible cost of running these processes and examine the employee time and attention they consume.
Speaking to People Matters, Wheeldon describes this as a broader productivity challenge, particularly when routine administration interrupts work requiring judgement, creativity, collaboration and customer engagement.
"A five-minute task repeated across thousands of employees can become a material productivity cost, particularly when it interrupts higher-value work," he says.
The cost goes beyond the time spent on a task
Organisations often measure administration through the cost of the team managing a process or the time needed to complete an individual task. Wheeldon says this overlooks the wider impact on employees and business operations.
He recommends establishing a baseline across the full employee journey, using measures such as:
Elapsed cycle time and active handling time
Manual touchpoints, hand-offs and approval delays
Rework and rejection rates
Cost per transaction and employee queries
Time taken to complete reimbursements
These measures should also be examined across business units, geographies and process types. The aim is to turn administrative burden from an anecdotal complaint into something organisations can track and address.
Wheeldon points to UST as an example of process redesign. Following changes informed by employee feedback and a lighter approval model, the company reports that 95% of domestic travel is now auto-approved, 67% of overall travel and expense spend has shifted to cards, and adoption of SAP Concur for travel bookings, expenses and cash advances has reached 100%.
The figures illustrate how process changes can affect employee effort while improving organisational visibility and control.
Digitising a process is not the same as simplifying it
The everyday sources of workplace friction are often unremarkable: finding policy-compliant travel options, entering information repeatedly, itemising receipts, checking reimbursement eligibility and following up on approvals.
Each task may appear manageable in isolation. Together, they create what Wheeldon calls a recurring "friction tax" on employees and managers.
He identifies a common weakness in digital transformation: organisations have digitised individual steps without redesigning the complete process. An online form may replace a paper form, but the approval chain, underlying policy and disconnected data remain unchanged.
"Digital transformation should therefore be judged by effort removed, not applications deployed," he says.
AI can help reduce some of this friction when it is built into the workflow. Wheeldon cites ExpenseIt, which uses optical character recognition, machine learning, generative AI and agentic AI to convert receipts into expense entries.
SAP Concur reports that the tool can itemise complex hotel bills in less than 30 seconds. Its Joule-enabled experiences also support natural-language interactions across travel and expense tasks.
When controls start slowing down decisions
Approvals and reporting requirements are necessary for financial control and compliance. However, applying the same approval process to every transaction can create delays without necessarily improving oversight.
Wheeldon says excessive controls can frustrate employees and encourage workarounds. Managers, meanwhile, spend time reviewing routine activity instead of focusing on coaching, customer needs or genuine exceptions.
There is also a decision-making cost. Manual data collection and reconciliation can leave leaders working with incomplete or outdated information, while rework delays reimbursements and weakens trust in internal processes.
He advocates a risk-based approach: routine, policy-compliant transactions should move efficiently, while human attention should focus on unusual patterns, exceptions and higher-risk activity.
AI tools such as SAP Concur's Verify and Intelligent Audit can support receipt checks, identify potential errors and flag policy issues for further review. The objective is more consistent controls without requiring every transaction to pass through the same level of manual scrutiny.
The danger of automating complexity
As organisations invest in AI, Wheeldon sees a risk in applying new technology to processes that have not been properly simplified.
An inefficient workflow may become faster at one step while remaining frustrating for employees overall. Disconnected data, unclear policies and excessive approvals can continue to undermine the experience.
"The right starting point is the outcome: what work should disappear, what decisions should improve, and where must human judgement remain?" he says.
He suggests simplifying processes and connecting data before introducing AI to anticipate needs, provide guidance or complete routine actions. Success should be assessed through fewer manual interactions, faster cycle times, reduced rework, stronger compliance and an improved employee experience.
SAP Concur is extending its AI capabilities through agent-based tools. Its Expense Automation Agent can create expense reports, add transactions and populate custom fields using contextual details and user history. Employees can then review and refine the report before submission.
The Expense Pre-Submit Audit Agent is designed to flag receipt discrepancies earlier, helping reduce inaccuracies, rejected reports and review cycles. Wheeldon stresses the importance of keeping these systems policy-aware, transparent and subject to appropriate human oversight.
Where should organisations draw the line?
For Wheeldon, the distinction between governance and bureaucracy comes down to whether a control manages a clearly defined risk.
Necessary governance has an accountable owner and demonstrable purpose. Bureaucracy, by contrast, adds steps even when they do not change the outcome.
He suggests leaders examine every approval or reporting requirement through five questions:
What risk is the control intended to manage?
How often does it identify a genuine exception?
Could the same risk be managed automatically?
Does every transaction require the same control?
How much employee effort and delay does it create?
Controls that generate few useful exceptions while consuming significant handling time may need redesign.
Wheeldon returns to UST's experience to illustrate how automation and governance can coexist. Its reported 95% auto-approval rate for domestic travel reflects a lighter process, while the organisation retains visibility into travel spending.
"That is the objective: fewer layers for employees, not less governance for the enterprise," he says.
Measuring the work organisations give back
AI's role in administration is not limited to completing tasks. It can also bring compliance checks earlier into a process by capturing receipt data, identifying missing information, explaining policies and flagging potential anomalies before submission.
Within SAP Concur, ExpenseIt supports receipt capture and expense creation, while Verify and Intelligent Audit apply AI and machine learning to receipt and policy checks. Joule offers conversational access to travel, expense and policy tasks.
Wheeldon also points to findings from SAP Concur's 2025 Global Business Travel Survey, which found 88% of business travellers globally were comfortable using AI-powered automation for at least some elements of business travel.
However, he says automation needs clear boundaries. Enterprise AI must protect data, preserve audit trails, operate within defined policies and identify when human intervention is required.
"The goal is not to remove people from consequential decisions. It is to remove repetitive administration so that people can focus on judgement, exceptions and outcomes," he says.
Looking ahead, Wheeldon expects organisations to measure process adherence alongside the effort employees spend navigating internal systems.
Compliance, transaction costs and risk events will remain important measures. But they need to sit alongside active handling time, manual touchpoints, first-time-right rates, approval delays and the time returned to employees.
He also cautions against measuring automation only by the work removed from one team. Organisations need to establish whether technology reduces effort across the entire process or simply shifts administration to managers and shared services.
For business and HR leaders, the larger question is therefore not how many AI features an organisation has deployed. It is whether employees have fewer repetitive tasks, leaders can make better-informed decisions and essential controls remain effective.
The productivity gains from AI may ultimately be measured not by the work it adds, but by the unnecessary work it takes away.
